• The AI Plug
  • Posts
  • #62 AI chatbot trainers are being exploited

#62 AI chatbot trainers are being exploited

Plus: News anchors beware

Sorry for the missed email last week. Back at it this Monday with some mind-boggling AI numbers and stories.
 

In today’s newsletter:

  • Hottest stories: AI’s underclass, news anchors beware, and the SEC proposes a crackdown

  • TL;DR Rundown: $900k job at Netflix, English is the cheapest AI language, and Alexa bringing AI to over 50 million homes

  • Tool of the day: Cliffnotes for your podcasts

  • Shower thoughts: Brain food for shower time

HOTTEST STORIES 
Today’s biggest stories if you’re in a rush 

Tech giants like OpenAI, Google, and Anthropic are use reinforcement learning with human feedback (RLHF) to sharpen their AI tech. The process is simple: flood a chatbot with texts, then bring in human touch to finesse it. But here's the rub - AI raters have it rough. Meager pay, vague instructions, and a ticking clock define their reality. Yet, despite their key role in the tech surge, they're the unsung heroes in the AI narrative. Looks like HR's future might have a few bots up its sleeve.

- - - - - - - - - - - - - -

India's joined the AI anchor club, debuting hyper-realistic bot-hosts that could pass for flesh and blood. Enter Sana, the AI brainchild rocking India Today Group's Aaj Tak news channel, delivering multilingual updates daily. As these bots invade the airwaves, human anchors sweat over job security. Granted, AI hosts stumble with monotonous tones, wonky blinks and stutters, but they ace repetitive and data-crunching tasks. Buckle up, South Asian broadcasting's on a wild tech rollercoaster ride.

- - - - - - - - - - - - -

According to the Washington Post, the SEC's putting the brakes on investment firms using AI to score more business at clients' expense. In their crosshairs? Online brokerages like Robinhood, known for using AI to influence customer trading. The SEC's proposed rule extends conflict-of-interest bans to AI-driven personal data exploitation. Although advisers must already prioritize clients' best interests, these rules crank up the stakes for online platforms. Firms would need to spotlight and nix any AI-induced conflicts of interest. The SEC's move could be a game changer for Wall Street's tech future.

GIVE US A CLICK AND TAKE YOUR PICK  
A gentleman’s agreement 

TL;DR RUNDOWN
Listicle of what else is happening today 

Netflix rich: Amid ongoing strikes from actors and writers, Netflix is hiring an AI manager. Yep, it’s a 6-figure salary. Actually, it’s a $900k figure salary, which would put “AI Manager @ Netflix” as a better-paid position than most CEO’s out there.

AI language inequality survivors: Apparently AI usage fees are up to 15x cheaper for English than other languages. Rest assured we’ll be hearing about this in the mainstream soon too.

Meta Ali: Alibaba’s cloud unit just brought Meta’s AI model Llama to its client base. A synergy of Western and Eastern powers if there ever was one.

Home AI: Amazon’s Alexa resides in over 50 million US homes, but Alexa’s foundational technology is far behind the new wave of AI flooding the market. Amazon aims to change that, by upgrading the Alexa software to the equivalent of a voice-enabled ChatGPT.

AutogenAI: A generative tool for writing bids and pitches secures $22.3M in funding, perpetuating the cycle of massive amounts of money being dropped into AI startups.

Night-mode activated: Bats echolocate to see in the dark. Dogs smell. And as mum and dad told you, eating carrots will help you see in the dark too. Well, so will this AI-enhanced night-vision technology.

AI surveillance: NYC’s subway system has joined airports and police stations in using AI-powered surveillance. Your morning and afternoon commutes are no longer your own, it seems.

Serious drones: “Necessity is the mother of invention,” as the saying goes. That couldn’t be more relevant than in Ukraine, where the war is spurring some serious AI drone firepower.

Mission not-so-likely: The new Mission Impossible film featured the Entity. Luckily, according to (this particular) expert, it’s not a realistic representation of what an all-destroying AI would be like — pure opinion.

TOOL OF THE DAY 
AI tools we’ve used, loved, and recommend above all others 

Today’s tool is dstill.ai.

Put simply, it’s cliff notes for podcast episodes. Some podcasts you can listen to for hours, no matter what - the hosts are great.

Others not so much - they’re whiny or think they’re funny but they aren’t - but they still talk about great topics and cover content you’re interested in.

That’s where dstill.ai comes in - get the skinny on any podcast episode you CBA listening to in full.

MONDAY SHOWER THOUGHTS  
Ponders from Reddit

  1. If humans ever go to war with AI, we’ll have to send coded messages to each other in the form of CAPTCHAS. LINK

  2. The jobs that AI might be best suited to replace are executive-level positions. LINK

  3. Eventually, we’ll be morally obligated to give rights to AI. LINK

  4. Billionaires are probably warning us abut AIs because the first thing any sentient AI would do is warn us about Billionaires. LINK

  5. We are going to need a security question between friends/family to make sure we aren’t speaking to an AI soon. LINK

Could You Help Us Out?
Share with one AI-curious friend and receive our in-depth prompt guide. Use this link

How’d we go today? Telling us makes a huge difference, it really does.